mrkurt

55 ms·

631.
by mrkurt 6y ago
If they sell zero, YC's investment is worth $0 just like it is for half of the companies they invest in. The success rate for hope-to-be-unicorn startups is very low.
632.
by mrkurt 6y ago
Zapier is basically the biggest "no code" environment on the planet. The people who get the most value out of them don't care about per task prices. They've turned several million people into productive developers.
633.
by mrkurt 6y ago
They're different things. NAT64 is useful for bridging between ipv6 and ipv4 networks, it's just a way of encoding ipv4 addresses in ipv6 addresses. It's more straightforward than ipv4 nat because there's no port rewriti
634.
by mrkurt 6y ago
SSRF is a tremendous problem with webhooks. Stripe also solves this: https://fly.io/blog/practical-smokescreen-sanitizing-your-ou...
635.
by mrkurt 6y ago
For what it's worth, Stripe's Idempotent Requests are a nice way to handle this: https://stripe.com/docs/api/idempotent_requests We just send the same requests over and over until they're successful
636.
by mrkurt 6y ago
Hey thanks! If you want to run Diahook on it we can help. We happen to be pretty good at infra, so a library / self hosted setup is usually preferable. This is partially because we prefer to monitor things ourselves, and partially beca
637.
by mrkurt 6y ago
This is great. I've wanted a library for this, but a SaaS will do. :)
638.
by mrkurt 6y ago
The source is about it, it's pretty readable though. If you're turning up microvms with a linux kernel, it might just be easier to use kernel mode wireguard. It works pretty well!
639.
by mrkurt 6y ago
It'll work through NAT for sure. We just don't have to think about NAT because we're connecting clients to a network we control. Tailscale is making a mesh across clients behind different NATs. I know they did a bunch of work
640.
by mrkurt 6y ago
I love proxies and think all problems should be solved with proxies. Which means – if you give Fly.io a try and need any help, you should let me know!
641.
by mrkurt 6y ago
"flyctl ssh issue" will get you all setup for normal ssh access, and even store credentials in an agent if you have one running.
642.
by mrkurt 6y ago
To be fair, what Tailscale is doing is much harder than our private networking. They have to deal with NAT, mobile OSes, etc. We mostly just try to pick the right primitives. And frequently get that wrong. Like that time we wrote our own JS
643.
by mrkurt 6y ago
That's how it works now! You just have to setup wireguard first. You don't need to compile anything. This userland wireguard project was helpful for making "flyctl run console" work.
644.
by mrkurt 6y ago
It should be an HN thing to say. Unrestrained positivity is much more fun than kneejerk cynicism. :D
645.
by mrkurt 6y ago
The link to the source code is in the middle of the article, jeez: https://github.com/superfly/flyctl/pull/368
646.
by mrkurt 6y ago
We are a hosting company. Customer apps run in isolated private networks. We let them connect to these private networks with WireGuard. Customers _also_ want to do things like "launch a console", so we give them a mechanism for SS
647.
by mrkurt 6y ago
Fly.io | Elixir Dev Advocate | Remote (US time-zones preferred) | Full Time Fly.io is a hosting platform for distributed applications. Our users give us containers; we transmute them into fleets of Firecracker micro-VMs and run them on a Wi
648.
by mrkurt 6y ago
We use (and abuse) Nomad for Fly.io workloads. My favorite thing about it is that it's simple to understand. You can read the whole project and pretty much figure out what's going on . We've written: * A custom task driver to
649.
by mrkurt 6y ago
Me? It's handy to be able to dump -> modify -> load jobs when I'm poking at things. No one said anything about serious work, though, they said "load json natively".
650.
by mrkurt 6y ago
It can only load json with the api, the nomad CLI can't launch json jobs.
651.
Fly.io (YC W20) Is Hiring Elixir Developer Advocates (Remote)
1 points by mrkurt 6y ago
652.
Persistent storage and fast remote builds(fly.io)
11 points by mrkurt 6y ago|0 comments
653.
by mrkurt 6y ago
We have dedicated hardware in a bunch of facilities, either leased or colo.
654.
by mrkurt 6y ago
Cockroach has a neat and almost transparent way of doing this. If you key your tables by customer, it'll happily move those chunks of data to the nodes that need them most often. We almost shipped cockroach but it's missing some P
655.
by mrkurt 6y ago
App servers should have a fast access to data! Which either means running caches alongside the app servers, or doing clever things with databases. We just released a preview Postgres + regional replicas. This setup keeps the postgres leader
656.
by mrkurt 6y ago
Shameless plug: https://fly.io We built Fly _specifically_ so you can run servers close to your users. We do a lot for you on the network side, too, like terminate TLS in all our regions. One thing to note, though, is that laten
657.
by mrkurt 6y ago
Firecracker snapshots to device mapper could be pretty cool. We don't do much with snapshot/restore but we do a _lot_ with device mapper. It's pretty badass.
658.
by mrkurt 6y ago
This is super cool. I would like to know more about VM hibernation!
659.
by mrkurt 6y ago
No it was a good question! I did not make that at all clear, and the distinction is important.
660.
by mrkurt 6y ago
Probably writable blu ray.

More ›